Towards building a trustworthy pipeline integrating Neuroscience Gateway and Open Science Chain.

When the scientific dataset evolves or is reused in workflows creating derived datasets, the integrity of the dataset with its metadata information, including provenance, needs to be securely preserved while providing assurances that they are not accidentally or maliciously altered during the process. Providing a secure method to efficiently share and verify the data as well as metadata is essential for the reuse of the scientific data. The National Science Foundation (NSF) funded Open Science Chain (OSC) utilizes consortium blockchain to provide a cyberinfrastructure solution to maintain integrity of the provenance metadata for published datasets and provides a way to perform independent verification of the dataset while promoting reuse and reproducibility. The NSF- and National Institutes of Health (NIH)-funded Neuroscience Gateway (NSG) provides a freely available web portal that allows neuroscience researchers to execute computational data analysis pipeline on high performance computing resources. Combined, the OSC and NSG platforms form an efficient, integrated framework to automatically and securely preserve and verify the integrity of the artifacts used in research workflows while using the NSG platform. This paper presents the results of the first study that integrates OSC-NSG frameworks to track the provenance of neurophysiological signal data analysis to study brain network dynamics using the Neuro-Integrative Connectivity tool, which is deployed in the NSG platform. Database URL: https://www.opensciencechain.org.

Evolutionary algorithm optimization of biological learning parameters in a biomimetic neuroprosthesis.

Biomimetic simulation permits neuroscientists to better understand the complex neuronal dynamics of the brain. Embedding a biomimetic simulation in a closed-loop neuroprosthesis, which can read and write signals from the brain, will permit applications for amelioration of motor, psychiatric, and memory-related brain disorders. Biomimetic neuroprostheses require real-time adaptation to changes in the external environment, thus constituting an example of a dynamic data-driven application system. As model fidelity increases, so does the number of parameters and the complexity of finding appropriate parameter configurations. Instead of adapting synaptic weights via machine learning, we employed major biological learning methods: spike-timing dependent plasticity and reinforcement learning. We optimized the learning metaparameters using evolutionary algorithms, which were implemented in parallel and which used an island model approach to obtain sufficient speed. We employed these methods to train a cortical spiking model to utilize macaque brain activity, indicating a selected target, to drive a virtual musculoskeletal arm with realistic anatomical and biomechanical properties to reach to that target. The optimized system was able to reproduce macaque data from a comparable experimental motor task. These techniques can be used to efficiently tune the parameters of multiscale systems, linking realistic neuronal dynamics to behavior, and thus providing a useful tool for neuroscience and neuroprosthetics.

Tau and amyloid-related pathologies in the entorhinal cortex have divergent effects in the hippocampal circuit.

The entorhinal cortex (EC) is affected early in Alzheimer's disease, an illness defined by a co-occurrence of tau and amyloid-related pathologies. How the co-occurrence of these pathologies in the EC affects the hippocampal circuit remains unknown. Here we address this question by performing electrophysiological analyses of the EC circuit in mice that express mutant human amyloid precursor protein (hAPP) or tau (hTau), or both in the EC. We show that the alterations in the hippocampal circuit are divergent, with hAPP increasing but hTau decreasing neuronal/circuit excitability. Most importantly, mice co-expressing hAPP and hTau show that hTau has a dominant effect, dampening the excitatory effects of hAPP. Additionally, compensatory synaptic downscaling, in response to increased excitability in EC was observed in subicular neurons of hAPP mice. Based on simulations, we propose that EC interneuron pruning can account for both EC hyperexcitability and subicular synaptic downscaling found in mice expressing hAPP.

Calcium regulation of HCN channels supports persistent activity in a multiscale model of neocortex.

Neuronal persistent activity has been primarily assessed in terms of electrical mechanisms, without attention to the complex array of molecular events that also control cell excitability. We developed a multiscale neocortical model proceeding from the molecular to the network level to assess the contributions of calcium (Ca(2+)) regulation of hyperpolarization-activated cyclic nucleotide-gated (HCN) channels in providing additional and complementary support of continuing activation in the network. The network contained 776 compartmental neurons arranged in the cortical layers, connected using synapses containing AMPA/NMDA/GABAA/GABAB receptors. Metabotropic glutamate receptors (mGluR) produced inositol triphosphate (IP3) which caused the release of Ca(2+) from endoplasmic reticulum (ER) stores, with reuptake by sarco/ER Ca(2+)-ATP-ase pumps (SERCA), and influence on HCN channels. Stimulus-induced depolarization led to Ca(2+) influx via NMDA and voltage-gated Ca(2+) channels (VGCCs). After a delay, mGluR activation led to ER Ca(2+) release via IP3 receptors. These factors increased HCN channel conductance and produced firing lasting for ∼1min. The model displayed inter-scale synergies among synaptic weights, excitation/inhibition balance, firing rates, membrane depolarization, Ca(2+) levels, regulation of HCN channels, and induction of persistent activity. The interaction between inhibition and Ca(2+) at the HCN channel nexus determined a limited range of inhibition strengths for which intracellular Ca(2+) could prepare population-specific persistent activity. Interactions between metabotropic and ionotropic inputs to the neuron demonstrated how multiple pathways could contribute in a complementary manner to persistent activity. Such redundancy and complementarity via multiple pathways is a critical feature of biological systems. Mediation of activation at different time scales, and through different pathways, would be expected to protect against disruption, in this case providing stability for persistent activity.

Local axon collaterals of area CA1 support spread of epileptiform discharges within CA1, but propagation is unidirectional.

CA3 and subiculum are hippocampal formation regions that can initiate seizure activity because each has a substantial intrinsic excitatory connectivity. We studied the intrinsic connectivity of area CA1 by exploring the spread of synchronous population discharges in ventral hippocampal slices from rats using a recording chamber that permitted multiple simultaneous extracellular recordings along all laminae of CA1. Brief single stimulus pulses were applied to stratum oriens (SO) or stratum radiatum (SR) on the CA3 side or the subicular side of CA1. In disinhibited slices, events triggered with SO or SR stimulation on the CA3-side propagated over the proximo-distal extent of CA1 with a maximal conduction velocity of 0.4 m/s, comparable with antidromic conduction velocities within CA1. By contrast, SO or SR stimuli applied on the subicular side of CA1 triggered events that did not spread "backward" toward CA3. These events are rapidly decremented in amplitude and duration. Whereas antidromic responses were largest when stimuli were applied on the subicular side of CA1, such responses were not sufficient to trigger epileptiform discharges when excitatory transmission was intact. We conclude that the unidirectional spread of epileptiform activity in area CA1 is the result of an intrinsic axon collateral system where each pyramidal cell has a proportionally larger projection toward subiculum. Although this collateral system is sparse compared with other hippocampal formation regions, its unidirectionality protects against re-entrant activation of CA3 and may be physiologically significant as a relay from proximal CA1 to distal CA1.

Parallel network simulations with NEURON.

The NEURON simulation environment has been extended to support parallel network simulations. Each processor integrates the equations for its subnet over an interval equal to the minimum (interprocessor) presynaptic spike generation to postsynaptic spike delivery connection delay. The performance of three published network models with very different spike patterns exhibits superlinear speedup on Beowulf clusters and demonstrates that spike communication overhead is often less than the benefit of an increased fraction of the entire problem fitting into high speed cache. On the EPFL IBM Blue Gene, almost linear speedup was obtained up to 100 processors. Increasing one model from 500 to 40,000 realistic cells exhibited almost linear speedup on 2,000 processors, with an integration time of 9.8 seconds and communication time of 1.3 seconds. The potential for speed-ups of several orders of magnitude makes practical the running of large network simulations that could otherwise not be explored.

Photic-induced sensitization: acquisition of an augmenting spike-wave response in the adult rat through repeated strobe exposure.

It is well established that patterns of sensory input can affect neuroplastic changes during early development. The scope and consequences of experience-dependent plasticity in the adult are less well understood. We studied the possibility that repeated exposure to trains of stroboscopic stimuli could induce a sensitized and potentially aberrant response in ordinary individuals. Chronic electrocorticographic recording electrodes enabled measurement of responses in awake, freely moving animals. Normal adult rats, primarily Sprague-Dawley, were exposed to 20-40 strobe trains per day after a strobe-free adaptation period. The common response to strobe trains changed in 34/36 rats with development of a high-amplitude spike-wave response that emerged fully by the third day of photic exposure. Onset of this sensitized response was marked by short-term augmentation of response to successive strobe flashes. The waveform generalized across the brain, reflected characteristics of the visual stimulus, as well as an inherent 6- to 8-Hz pacing, and was suppressed with ethosuximide administration. Spike-wave episodes were self-limiting but could persist beyond the strobe period. Sensitization lasted 2-4 wk after last strobe exposure. The results indicate visual stimulation, by itself, can induce in adult rats an enduring sensitization of visual response with epileptiform characteristics. The results raise the question of the effects of such neuroplastic change on sensation and epileptiform events.

Alternating dominance of NMDA and AMPA for learning and recall: a computer model.

Physiological studies reveal a dichotomy in biological Hebbian learning: NMDA receptors are utilized for induction of long term potentiation (LTP) whereas AMPA is used for LTP expression. We propose that this dichotomy would have functional value: preventing previously stored memories from interfering with the storage of new memories. A previous hypothesis reduces this interference by temporarily reducing associative weights during learning. Complementary to this model, we propose a dual transmission algorithm in which one set of synaptic weights are used primarily for learning and another primarily for recall. This algorithm shows good performance in a simple neural network model. Biologically, the model could be mediated by a cholinergic switch from dominance of learning-insensitive NMDA receptors to dominance of learning-modifiable AMPA receptors.

Can the hippocampus tell time? The temporo-septal engram shift model.

An essential feature of episodic memory, the type of memory dependent on hippocampus, is that individual memories belong to particular moments in time. Recent PET studies suggest that memory encoding and recall occur at different locations in human hippocampus. Coupled with other attributes of hippocampus, this suggested to us that the septo-temporal hippocampal axis may play an important role in time perception. We propose a temporo-septal engram shift model of hippocampal memory. The model posits that memories gradually move along the hippocampus from a temporal encoding site to ever more septal sites from which they are recalled. We propose that the sense of time is encoded by the location of the engram along the temporo-septal axis.

Properties of a hyperpolarization-activated cation current in interneurons in the rat lateral geniculate nucleus.

A hyperpolarization-activated cation conductance contributes to the membrane properties of a variety of cell types. In the thalamus, a prominent hyperpolarization-activated cation conductance exists in thalamocortical cells, and this current is implicated in the neuromodulation of complex firing behaviors. In contrast, the GABAergic cells in the reticular nucleus in the thalamus appear to lack this conductance. The presence and role of this cation conductance in the other type of thalamic GABAergic cells, local interneurons, is still unclear. To resolve this issue, we studied 54 physiologically and morphologically identified local interneurons in the rat dorsal lateral geniculate nucleus using an in vitro whole-cell patch recording technique. We found that hyperpolarizing current injections induced depolarizing voltage sags in these geniculate interneurons. The I-V relationship revealed an inward rectification. Voltage-clamp study indicated that a slow, hyperpolarization-activated cation conductance was responsible for the inward rectification. We then confirmed that this slow conductance had properties of the hyperpolarization-activated cation conductance described in other cell types. The slow conductance was insensitive to 10 mM tetraethylammonium and 0.5 mM 4-aminopyridine, but was largely blocked by 1-1.5 mM Cs+. It was permeable to both K+ and Na+ ions and had a reversal potential of -44 mV. The voltage dependence of the hyperpolarization-activated cation conductance in interneurons was also studied: the activation threshold was about -55 mV, half-activation potential was about -80 mV and maximal conductance was about 1 nS. The activation and deactivation time constants of the conductance ranged from 100 to 1000 ms, depending on membrane potential. The depolarizing voltage sags and I-V relationship were further simulated in a model interneuron, using the parameters of the hyperpolarization-activated cation conductance obtained from the voltage-clamp study. The time-course and voltage dependence of the depolarizing voltage sags and I-V relationship in the model cell were very similar to those found in geniculate interneurons in current clamp. Taken together, the results of the present study suggest that thalamic local interneurons possess a prominent hyperpolarization-activated cation conductance, which may play important roles in determining basic membrane properties and in modulating firing patterns.

Burst firing in identified rat geniculate interneurons.

We used whole-cell patch recording to study 102 local interneurons in the rat dorsal lateral geniculate nucleus in vitro. Input impedance with this technique (607.0+/-222.4 MOhm) was far larger than that measured with sharp electrode techniques, suggesting that interneurons may be more electrotonically compact than previously believed. Consistent and robust burst firing was observed in all interneurons when a slight depolarizing boost was given from a potential at, or slightly hyperpolarized from, resting membrane potential. These bursts had some similarities to the low-threshold spike described previously in other thalamic neuron types. The bursting responses were blocked by Ni+, suggesting that the low-threshold calcium current I(T), responsible for the low-threshold spike, was also involved in interneuron burst firing. Compared to the low-threshold spike of thalamocortical cells, however, the interneuron bursts were of relatively long duration and low intraburst frequency. The requirement for a depolarizing boost to elicit the burst is consistent with previous reports of a depolarizing shift of the I(T) activation curve of interneurons relative to thalamocortical cells, a finding we confirmed using voltage-clamp. Voltage-clamp study also revealed an additional long-lasting current that could be tentatively identified as the calcium activated non-selective cation current, I(CAN), based on reversal potential and on pharmacological characteristics. Computer simulation of the interneuron burst demonstrated that its particular morphology is likely due to the interaction of I(T) and I(CAN). In the slice, bursts could also be elicited by stimulation of the optic tract, suggesting that they may occur in response to natural stimulation. Synaptically triggered bursts were only partially blocked by Ni+, but could then be completely blocked by further addition of (+/-)-2-amino-5-phosphonopentanoic acid. The existence of robust bursts in this cell type suggests an additional role for interneurons in sculpting sensory responses by feedforward inhibition of thalamocortical cells. The low-threshold spike is a mechanism whereby activity in a neuron is dependent on a prior lack of activity in that same neuron. Understanding of the low-threshold spike in the other major neuron types of the thalamus has brought many new insights into how thalamic oscillations might be involved in sleep and epilepsy. Our description of this phenomenon in the interneurons of the thalamus suggests that these network oscillations might be even more complicated than previously believed.

An intrinsic oscillation in interneurons of the rat lateral geniculate nucleus.

By using the whole cell patch recording technique in vitro, we examined the voltage-dependent firing patterns of 69 interneurons in the rat dorsal lateral geniculate nucleus (LGN). When held at a hyperpolarized membrane potential, all interneurons responded with a burst of action potentials. In 48 interneurons, larger current pulses produced a bursting oscillation. When relatively depolarized, some interneurons produced a tonic train of action potentials in response to a depolarizing current pulse. However, most interneurons produced only oscillations, regardless of polarization level. The oscillation was insensitive to the bath application of a combination of blockers to excitatory and inhibitory synaptic transmission, including 30 microM 6,7-dinitroquinoxaline-2,3-dione, 100 microM (+/-)-2-amino-5-phosphonopentanoic acid, 20 microM bicuculline, and 2 mM saclofen, suggesting an intrinsic event. The frequency of the oscillation in interneurons was dependent on the intensity of the injection current. Increasing current intensity increased the oscillation frequency. The maximal frequency of the oscillation was 5-15 Hz for most cells, with some ambiguity caused by the difficulty of precisely defining a transition from oscillatory to regular firing behavior. In contrast, the interneuron oscillation was little affected by preceding depolarizing and hyperpolarizing pulses. In addition to being elicited by depolarizing current injections, the oscillation could also be initiated by electrical stimulation of the optic tract when the interneurons were held at a depolarized membrane potential. This suggests that interneurons may be recruited into thalamic oscillations by synaptic inputs. These results indicate that interneurons may play a larger role in thalamic oscillations than was previously thought.

Potentiation of Ca2+ influx through NMDA channels by action potentials: a computer model.

In pyramidal cells, somatic action potentials can propagate actively back into the apical dendrites and potentiate calcium influx at simultaneously activated glutamatergic synapses, presumably by relieving the voltage-dependent block of NMDA channels. We have used computer simulations to investigate the conditions under which this potentiation will be optimal. We find that a spike with a long duration and limited amplitude (peak of approximately -10 mV) will be most effective. A back-propagating action potential will achieve this form if the dendritic membrane has a low K+ channel density and a modest Na+ channel density (30-70 pS/microm2, similar to experimentally observed densities). The relative increase in calcium due to the backpropagating spike will be small, however, unless the accumulated calcium is rapidly removed.

Computer models of hippocampal circuit changes of the kindling model of epilepsy.

Abnormalities in the organization of brain circuits may underlie many types of epilepsy. This hypothesis can best be evaluated in the case of temporal lobe epilepsy, where evidence of rewiring (synaptic reorganization) can be found in the dentate gyrus. Computer modeling of normal and reorganized dentate gyrus was used to understand the functional consequences of these structural changes. Hyperexcitability appeared to be largely limited by the powerful intrinsic adaptation characteristic of granule cells, the principal cells in this area. Combining disinhibition with new recurrent excitatory circuitry was necessary to produce repeated firing of these cells. Paradoxically, continuing regenerative activity was only seen with a large reduction in the strength of the inciting stimulus. Validation of these findings will require further physiological correlation.

Computer model of antiepileptic effects mediated by alterations in GABA(A)-mediated inhibition.

Results from a computer model of a thalamic network predict that agents augmenting GABA(A)-mediated inhibition in the reticular thalamic (RE) nucleus will be antiepileptic or desynchronizing. This provides support for the hypothesis that antiepileptics like benzodiazepines may exert their effects through an isolated increase of inhibition in the RE nucleus. When desynchronized, the model thalamocortical neurons showed a decreased probability of firing a low threshold spike, a decreased secondary inhibitory postsynaptic potential and a higher frequency of oscillations. The transition to desynchrony was also accompanied by an increased frequency in the firing of the model RE neurons.

Adapting a feedforward heteroassociative network to Hodgkin-Huxley dynamics.

Using the original McCulloch-Pitts notion of simple on and off spike coding in lieu of rate coding, an Anderson-Kohonen artificial neural network (ANN) associative memory model was ported to a neuronal network with Hodgkin-Huxley dynamics. In the ANN, the use of 0/1 (no-spike/spike) units introduced a cross-talk term that had to be compensated by introducing balanced feedforward inhibition. The resulting ANN showed good capacity and fair selectivity (rejection of unknown input vectors). Translation to the Hodgkin-Huxley model resulted in a network that was functional but not at all robust. Evaluation of the weaknesses of this network revealed that it functioned far better using spike timing, rather than spike occurrence, as the code. The algorithm requires a novel learning algorithm for feedforward inhibition that could be sought physiologically.

Computer model of clonazepam's effect in thalamic slice.

In the thalamus, paradoxical changes in response to augmentation of inhibition can occur as a result of either cellular or network effects. Clonazepam, a GABA(A) agonist, produces a paradoxical reduction in evoked thalamocortical neuron inhibitory postsynaptic potential (IPSP) in thalamic slice. This has been hypothesized to be a result of augmentation in inhibitory to inhibitory connections. In a computer model, orthodromic simulation produced an increase in initial IPSP, a result contrary to that found experimentally. This failure was traced to the inability of orthodromic activation to produce fast enough recurrent inhibition to alter initial reticularis neuron firing. Simulated antidromic stimulation was able to reduce this initial spike train and reproduced the experimental finding.

GABAA-mediated IPSCs in piriform cortex have fast and slow components with different properties and locations on pyramidal cells.

GABAA-mediated IPSCs in piriform cortex have fast and slow components with different properties and locations on pyramidal cells. J. Neurophysiol. 78: 2531-2545, 1997. A recent study in piriform (olfactory) cortex provided evidence that, as in hippocampus and neocortex, gamma-aminobutyric acid-A (GABAA)-mediated inhibition is generated in dendrites of pyramidal cells, not just in the somatic region as previously believed. This study examines selected properties of GABAA inhibitory postsynaptic currents (IPSCs) in dendritic and somatic regions that could provide insight into their functional roles. Pharmacologically isolated GABAA-mediated IPSCs were studied by whole cell patch recording in slices. To compare properties of IPSCs in distal dendritic and somatic regions, local stimulation was carried out with tungsten microelectrodes, and spatially restricted blockade of GABAA-mediated inhibition was achieved by pressure-ejection of bicuculline from micropipettes. The results revealed that largely independent circuits generate GABAA inhibition in distal apical dendritic and somatic regions. With such independence, a selective decrease in dendritic-region inhibition could enhance integrative or plastic processes in dendrites while allowing feedback inhibition in the somatic region to restrain system excitability. This could allow modulatory fiber systems from the basal forebrain or brain stem, for example, to change the functional state of the cortex by altering the excitability of interneurons that mediate dendritic inhibition without increasing the propensity for regenerative bursting in this highly epileptogenic system. As in hippocampus, GABAA-mediated IPSCs were found to have fast and slow components with time constants of decay on the order of 10 and 40 ms, respectively, at 29 degrees C. Modeling analysis supported physiological evidence that the slow time constant represents a true IPSC component rather than an artifactual slowing of the fast component from voltage clamp of a dendritic current. The results indicated that, whereas both dendritic and somatic-region IPSCs have both fast and slow GABAA components, there is a greater proportion of the slow component in dendrites. In a companion paper, the hypothesis is explored that the resulting slower time course of the dendritic IPSC increases its capacity to regulate the N-methyl--aspartate component of EPSPs. Finally, evidence is presented that the slow GABAA-mediated IPSC component is regulated by presynaptic GABAB inhibition whereas the fast is not. Based on the requirement for presynaptic GABAB-mediated block of inhibition for expression of long-term potentiation, this finding is consistent with participation of the slow GABAA component in regulation of synaptic plasticity. The lack of susceptibility of the fast GABAA component to the long-lasting, activity-induced suppression mediated by presynaptic GABAB receptors is consistent with a protective role for this process in preventing seizure activity.

Regulation of the NMDA component of EPSPs by different components of postsynaptic GABAergic inhibition: computer simulation analysis in piriform cortex.

Regulation of the NMDA component of EPSPs by different components of postsynaptic GABAergic inhibition: computer simulation analysis in piriform cortex. J. Neurophysiol. 78: 2546-2559, 1997. Physiological analysis in the companion paper demonstrated that gamma-aminobutyric acid-A (GABAA)-mediated inhibition in piriform cortex is generated by circuits that are largely independent in apical dendritic and somatic regions of pyramidal cells and that GABAA-mediated inhibitory postsynaptic currents (IPSCs) in distal dendrites have a slower time course than those in the somatic region. This study used modeling methods to explore these characteristics of GABAA-mediated inhibition with respect to regulation of the N-methyl--aspartate (NMDA) component of excitatory postsynaptic potentials. Such regulation is relevant to understanding NMDA-dependent long-term potentiation (LTP) and the integration of repetitive synaptic inputs that can activate the NMDA component as well as pathological processes that can be activated by overexpression of the NMDA component. A working hypothesis was that the independence and differing properties of IPSCs in apical dendritic and somatic regions provide a means whereby the NMDA component and other dendritic processes can be controlled by way of GABAergic tone without substantially altering system excitability. The analysis was performed on a branched compartmental model of a pyramidal cell in piriform cortex constructed with physiological and anatomic data derived by whole cell patch recording. Simulations with the model revealed that NMDA expression is more effectively blocked by the slow GABAA component than the fast. Because the slow component is present in greater proportion in apical dendritic than somatic regions, this characteristic would increase the capacity of dendritic IPSCs to regulate NMDA-mediated processes. The simulations further revealed that somatic-region GABAergic inhibition can regulate the generation of action potentials with little effect on the NMDA component generated by afferent fibers in apical dendrites. As a result, if expression of the NMDA component or other dendritic processes were enabled by selective block of dendritic inhibition, for example, by centrifugal fiber systems that may regulate learning and memory, the somatic-region IPSC could preserve system stability through feedback regulation of firing without counteracting the effect of the dendritic-region block. Simulations with paired inputs revealed that the dendritic GABAA-mediated IPSC can regulate the extent to which a strong excitatory input facilitates the NMDA component of a concurrent weak input, providing a possible mechanism for control of "associative LTP" that has been demonstrated in this system. Postsynaptic GABAB-mediated inhibition had less effect on the NMDA component than either the fast or slow GABAA components. Depolarization from a concomitant alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) component also was found to have comparatively little effect on current through the NMDA channel because of its brief time course.